1. Field of the Invention
This invention relates to a text voice readup system for converting text data into voices and reading up the voices.
2. Description of the Related Art
FIG. 2 is a block diagram showing the configuration of a conventional text voice readup system. FIG. 3 is a flowchart for describing a process of outputting a voice waveform from the text voice readup system shown in FIG. 2.
The text voice readup system comprises a host device 10 and a text readup device 20. The host device 10 has a readup controller 11. The readup controller 11 supplies text data through its output to the text readup device 20. The text readup device 20 comprises an analysis controller 21, a read-only memory (hereinafter called "ROM") 22 having a word dictionary and phonemic piece data stored therein, and a speaker 23. The analysis controller 21 is composed of a CPU or the like, for example. The ROM 22 is electrically connected to the analysis controller 21. Namely, the analysis controller 21 and the ROM 22 are composed of programmable hardware. The text data is supplied to an input port of the analysis controller 21 from which it is outputted as a voice waveform through a voice output port thereof, followed by input to the speaker 23.
When a readup request and its contents or the like are passed to the readup controller 11 of the host device 10, the readup controller 11 creates a text to generate text data. When the text data is transferred to the analysis controller 21 in the text readup device 20, the analysis controller 21 performs a text analytical process S1 by reference to the word dictionary stored in the ROM 22 to create an intermediate language. When, for example, text data "Turn to the right at a Koyama intersection located 100 m ahead" is input, the analysis controller 21 retrieves the word dictionary to check the way of reading the data and its accent, thereby creating an intermediate language "Hyakumetorusakino koyamakousatenwo miginimagatsutekudasai". In accordance with the subsequent composite or synthetic parameter process S2, the analysis controller 21 uses the phonemic piece data with respect to the intermediate language to generate a synthetic parameter and converts it into a voice waveform in accordance with a voice synthesis process S3. The voice waveform is inputted to the speaker 23 from which a voice corresponding to the text data is sounded or produced. Thus, the system shown in FIG. 2 is operated based on software programs to thereby read up the text voice.
However, the conventional text voice readup system has the following problems.
When one attempts to store all place names in the ROM 22 in the text voice readup system such as a place-name readup car navigation system or the like, the following problems (1) and (2) arise:
(1) Data about a dictionary with for place names held in a dictionary becomes massive (e.g., 1.5 MB for 100,000 words) and hence the speed of retrieving the data from retreiving the ROM 22 is reduced. Further, since the capacity of the ROM 22 must increase, its cost is raised beyond the need. PA1 (2) When homo-allophonic words or homographs which are read differently in the same notation exist, they cannot be suitably read up. PA1 area gazetteer reading means for reading an area gazetteer corresponding to information about the present position of a vehicle from divided area gazetteers; PA1 an analysis controller for generating text data corresponding to the present position information and analyzing the text data, using the area gazetteer as a high-priority temporary dictionary; and PA1 a text readup device for producing voices corresponding to the text data.